Publications by authors named "Renan Alves de Oliveira"
Article Synopsis
- A neural network model was developed to predict the evolution of cosmological -body simulations, suggesting it effectively approximates the Green's function relating initial conditions to outcomes in complex, nonlinear scenarios.
- The model was tested against various simpler configurations, demonstrating good generalization and the ability to learn physical principles from random Gaussian training data, which highlighted its strengths and weaknesses.
- However, the model struggled with initial conditions that it wasn't trained on, specifically those involving transverse modes, and it achieved high accuracy when compared to traditional simulation methods for density and momentum power spectra, outperforming a fast simulation method called COLA.
